In the information processing paradigm, long-term semantic memory or knowledge may be said to constitute the underlying structure of perceptual, cognitive and memory functions in each cerebral hemisphere. Four experiments are proposed to explore hemispheric differences in long-term semantic memory or knowledge. In the first experiment, I study the effect of presence (organized), absence (disorganized), or simple violation (incongruous) of familiar semantic organization on memory for pictures depicting multiple-object scenes. In the second experiment, I study hemispheric differences in specific semantic relations. The stimuli are pictures depicting single, common objects which vary in degree of representativeness (typicality) of natural superordinate categories, and the task requires decision as to whether two pictorial instances belong to a designated category. The third experiment is a natural sequel to the second one. In it the pictorial instances of categories used in the second experiment are replaced by words. In the last experiment, a finer analysis of hemispheric differences in coneptual semantic organization is undertaken in a feature-concept matching task that requires the subject to decide whether or not a flashed word or picture (the feature, e.g., "robin") describes another word presented in dichotic listening (the concept, e.g., "bird"). In order to obtain converging evidence from pathology and positive competence, experiments are administered to hemisphere-damaged stroke patients (experiment I) and to normal subjects (experiments I-IV). Neurological patients are tested in free vision and normal subjects in either a hemi-field tachistoscopic paradigm (experiments I-III), or a combined tachistoscopic/dichotic listening paradigm (experiment IV). All of the tests require non-verbal responses in order to remove a bias in favor of the left, speaking hemisphere.